The ability to design learning assessments is an essential competence for teachers to measure students’ achievement, evaluate the effectiveness of the learning process, and provide constructive feedback. This community service program was carried out to enhance the competence of elementary school teachers in the Suela Cluster, East Lombok, in developing relevant, interactive mathematics assessments supported by Artificial Intelligence (AI). The implementation method consisted of five stages: (1) needs analysis through teacher interviews, (2) intensive training on the use of AI and supporting applications such as Canva Magic AI and Almanack, (3) workshops on interactive assessment design, (4) classroom implementation mentoring, and (5) focus group discussions (FGD) on implementation results. Data were collected using questionnaires completed by 15 teachers, covering 13 indicators that assessed clarity of material delivery, trainer mastery, relevance, participation, usefulness, and the impact of training on teaching practice. The results showed that the average score per indicator ranged from 3.60 to 3.87 (scale 1–4), with the highest scores in clarity of delivery and relevance, and the lowest in time effectiveness. Qualitative analysis emphasized the need for longer duration, more varied practice examples (especially for lower grades), deeper materials, and program sustainability. Thus, the training proved effective in improving elementary school teachers’ competence in understanding and utilizing AI for mathematics assessment, although optimization is still required in technical aspects and program continuity.